Development of a Prognostic Nomogram for Modified Tinnitus Relieving Sound Therapy for Subjective Tinnitus.

Abstract

We aimed to develop a modified tinnitus-relieving sound system and establish a model for predicting its treatment effects. Retrospective study. Tinnitus Specialist Clinic of Eye & ENT Hospital, Fudan University. We recruited 107 patients undergoing modified tinnitus-relieving sounds between August 2020 and May 2021. Patients were divided into training (n = 75) and validation (n = 32) cohorts in a 7:3 ratio. The treatment outcome was Tinnitus Handicapped Inventory scores. Features were established using a least absolute shrinkage and selection operator-derived logistic regression model, where the selected clinical risk factors were included in the multivariate logistic regression, and a nomogram was established based on the model. The discrimination and calibration abilities of the nomogram were evaluated using the Hosmer-Lemeshow test and calibration curves. Decision curve analysis (DCA) was used to evaluate the net benefit of predictive efficacy. Multivariate logistic analysis indicated that the initial Tinnitus Handicapped Inventory score (odds ratio [OR]: 1.13 [1.07-1.19], P < .001) and treatment duration (OR: 3.4 [1.34-8.62], P < .001) were positive factors for improved tinnitus. The nomogram model that included baseline Tinnitus Handicapped Inventory score and treatment duration achieved a better concordance index of 0.880. DCA revealed that the nomogram model could lead to net benefits and exhibited a wider range of threshold probabilities for the prediction of therapeutic effects. Our study suggests that the nomogram model, including baseline Tinnitus Handicapped Inventory score and treatment duration, could achieve optimal performance in the preoperative prediction of the therapeutic effect of modified tinnitus-relieving sound.